pgpool2.git
9 years agoTranslate advanced.sgml into Japanese.
Tatsuo Ishii [Tue, 25 Oct 2016 01:13:27 +0000 (10:13 +0900)]
Translate advanced.sgml into Japanese.

Also fix typos in the English doc.

9 years agoTranslate runtime.sgml into Japanese
Yugo Nagata [Mon, 24 Oct 2016 11:47:06 +0000 (20:47 +0900)]
Translate runtime.sgml into Japanese

9 years agoTranslate installation.sgml into Japanese
Yugo Nagata [Mon, 24 Oct 2016 08:06:55 +0000 (17:06 +0900)]
Translate installation.sgml into Japanese

9 years agoTranslate watchdog.sgml into Japanese.
Tatsuo Ishii [Mon, 24 Oct 2016 04:45:33 +0000 (13:45 +0900)]
Translate watchdog.sgml into Japanese.

9 years agoTranslate figure title.
Tatsuo Ishii [Mon, 24 Oct 2016 01:07:25 +0000 (10:07 +0900)]
Translate figure title.

9 years agoFix typos and add some tags
Yugo Nagata [Fri, 21 Oct 2016 08:30:27 +0000 (17:30 +0900)]
Fix typos and add some tags

9 years agoTranslate start.sgml into japanese.
Tatsuo Ishii [Fri, 21 Oct 2016 08:19:57 +0000 (17:19 +0900)]
Translate start.sgml into japanese.

9 years agoOriginal document in english was updated
Yugo Nagata [Fri, 21 Oct 2016 07:20:18 +0000 (16:20 +0900)]
Original document in english was updated

9 years agoReplace pgpool-II into Pgpool-II and add some tags
Yugo Nagata [Fri, 21 Oct 2016 07:19:10 +0000 (16:19 +0900)]
Replace pgpool-II into Pgpool-II and add some tags

9 years agoRemove description about bzip2 from installation.sgml
Yugo Nagata [Fri, 21 Oct 2016 06:45:46 +0000 (15:45 +0900)]
Remove description about bzip2 from installation.sgml

We don't provide .tar.bz2 packages

9 years agoFix a type of installation.sgml
Yugo Nagata [Fri, 21 Oct 2016 06:44:46 +0000 (15:44 +0900)]
Fix a type of installation.sgml

This was explanation about PostgreSQL not Pgpool-II

9 years agoFix a type of installation.sgml
Yugo Nagata [Fri, 21 Oct 2016 06:34:42 +0000 (15:34 +0900)]
Fix a type of installation.sgml

This was explanation about PostgreSQL not Pgpool-II

9 years agoRevert "Fix typo"
Yugo Nagata [Fri, 21 Oct 2016 06:30:27 +0000 (15:30 +0900)]
Revert "Fix typo"

Unnecessary html files were commited by mistake.

This reverts commit 8e86363d08fa67492491b86c489da5d992d62bfe.

9 years agoFix typo
Yugo Nagata [Fri, 21 Oct 2016 06:23:40 +0000 (15:23 +0900)]
Fix typo

This was explanation about PostgreSQL not Pgpool-II

9 years agoFix small typos and translate tutorial section.
Tatsuo Ishii [Fri, 21 Oct 2016 05:27:13 +0000 (14:27 +0900)]
Fix small typos and translate tutorial section.

9 years agoTranslate problems.sgml into Japanese.
Tatsuo Ishii [Fri, 21 Oct 2016 05:18:00 +0000 (14:18 +0900)]
Translate problems.sgml into Japanese.

9 years agoTranslate restrictions.sgml into Japanese.
Tatsuo Ishii [Fri, 21 Oct 2016 03:32:45 +0000 (12:32 +0900)]
Translate restrictions.sgml into Japanese.

Fix errors in restrictions.sgml (English).

9 years agoTranslate info.sgml into Japanese.
Tatsuo Ishii [Fri, 21 Oct 2016 01:16:30 +0000 (10:16 +0900)]
Translate info.sgml into Japanese.

9 years agoTranslate notation.sgml into Japanese.
Tatsuo Ishii [Fri, 21 Oct 2016 00:59:28 +0000 (09:59 +0900)]
Translate notation.sgml into Japanese.

9 years agoTranslate history.sgml into Japanese.
Tatsuo Ishii [Fri, 21 Oct 2016 00:48:39 +0000 (09:48 +0900)]
Translate history.sgml into Japanese.

9 years agoAdd "lang=ja" so that Japanese navigation links are generated.
Tatsuo Ishii [Thu, 20 Oct 2016 23:32:12 +0000 (08:32 +0900)]
Add "lang=ja" so that Japanese navigation links are generated.

9 years agoTranslate pgpool.sgml to Japanese about Server Administration
Yugo Nagata [Thu, 20 Oct 2016 05:50:59 +0000 (14:50 +0900)]
Translate pgpool.sgml to Japanese about Server Administration

9 years agoTranslate intro.sgml into Japanese.
Tatsuo Ishii [Wed, 19 Oct 2016 08:49:05 +0000 (17:49 +0900)]
Translate intro.sgml into Japanese.

Fix typo in English intro.sgml.

9 years agoFix for [pgpool-hackers: 1850] PGPOOL SET/RESET SGML manual
Muhammad Usama [Wed, 19 Oct 2016 07:19:26 +0000 (12:19 +0500)]
Fix for [pgpool-hackers: 1850] PGPOOL SET/RESET SGML manual

9 years agoAdd missing files etc. in the previous commit.
Tatsuo Ishii [Wed, 19 Oct 2016 04:51:00 +0000 (13:51 +0900)]
Add missing files etc. in the previous commit.

9 years agoThis is the starting point of Japanese doc translation work.
Tatsuo Ishii [Wed, 19 Oct 2016 04:41:34 +0000 (13:41 +0900)]
This is the starting point of Japanese doc translation work.

9 years agoPrepare 3.6 alpha1 V3_6_0_ALPHA1
pengbo [Wed, 19 Oct 2016 03:15:31 +0000 (12:15 +0900)]
Prepare 3.6 alpha1

9 years agofix some mistakes in sgml release note
pengbo [Wed, 19 Oct 2016 03:08:16 +0000 (12:08 +0900)]
fix some mistakes in sgml release note

9 years agoFix for 0000252: authentication failed error when PCP command is cancelled...
Muhammad Usama [Tue, 18 Oct 2016 10:46:15 +0000 (15:46 +0500)]
Fix for 0000252: authentication failed error when PCP command is cancelled...

Fixed by adding the volatile modifier to authenticated flag, to make sure we
always get the updated value after the long jump.

9 years agoAdd a process diagram to "Architectural Fundamentals"
Tatsuo Ishii [Tue, 18 Oct 2016 08:19:32 +0000 (17:19 +0900)]
Add a process diagram to "Architectural Fundamentals"

9 years agoadd Pgpool-3.6 release-note (part of changes and bug fix)
pengbo [Tue, 18 Oct 2016 07:36:03 +0000 (16:36 +0900)]
add Pgpool-3.6 release-note (part of changes and bug fix)

9 years agoAdd more explanation about load balancing.
Tatsuo Ishii [Tue, 18 Oct 2016 07:07:35 +0000 (16:07 +0900)]
Add more explanation about load balancing.

9 years agoFix typo.
Tatsuo Ishii [Tue, 18 Oct 2016 06:57:20 +0000 (15:57 +0900)]
Fix typo.

9 years agoAdd link to restriction section.
Tatsuo Ishii [Tue, 18 Oct 2016 01:17:32 +0000 (10:17 +0900)]
Add link to restriction section.

9 years agoAdd remaining pseudo SQL command manuals.
Tatsuo Ishii [Mon, 17 Oct 2016 06:15:52 +0000 (15:15 +0900)]
Add remaining pseudo SQL command manuals.

9 years agoAdd show pool_nodes, show pool_processes and show pool_status
Tatsuo Ishii [Mon, 17 Oct 2016 04:45:05 +0000 (13:45 +0900)]
Add show pool_nodes, show pool_processes and show pool_status
reference manuals.

9 years agoEnhancements part of Pgpool-II 3.6 release note.
Tatsuo Ishii [Mon, 17 Oct 2016 03:50:58 +0000 (12:50 +0900)]
Enhancements part of Pgpool-II 3.6 release note.

9 years agoAdding SGML docs for PGPOOL SET/RESET/SHOW commands and Writing the setting parameter...
Muhammad Usama [Fri, 14 Oct 2016 21:14:25 +0000 (02:14 +0500)]
Adding SGML docs for PGPOOL SET/RESET/SHOW commands and Writing the setting parameters section

9 years agoAdding watchdog example to SGML doc
Muhammad Usama [Fri, 14 Oct 2016 13:14:20 +0000 (18:14 +0500)]
Adding watchdog example to SGML doc

9 years agoRelease note for 3.6 in progress.
Tatsuo Ishii [Fri, 14 Oct 2016 06:09:50 +0000 (15:09 +0900)]
Release note for 3.6 in progress.

9 years agoFix typo.
Tatsuo Ishii [Fri, 14 Oct 2016 02:52:19 +0000 (11:52 +0900)]
Fix typo.

9 years agoFix misc typos.
Tatsuo Ishii [Fri, 14 Oct 2016 02:47:22 +0000 (11:47 +0900)]
Fix misc typos.

9 years agoSGML Doc for Basic Configuration Example
Muhammad Usama [Thu, 13 Oct 2016 19:13:06 +0000 (00:13 +0500)]
SGML Doc for Basic Configuration Example

9 years agoSome more SGML docs
Muhammad Usama [Thu, 13 Oct 2016 14:51:55 +0000 (19:51 +0500)]
Some more SGML docs

9 years agoWriting the misc config parameters section of SGML docs
Muhammad Usama [Tue, 11 Oct 2016 12:23:10 +0000 (17:23 +0500)]
Writing the misc config parameters section of SGML docs

9 years agoAdd release notes.
Tatsuo Ishii [Tue, 11 Oct 2016 05:20:04 +0000 (14:20 +0900)]
Add release notes.

9 years agoUpdate Appendix A. Release Notes.
Tatsuo Ishii [Tue, 11 Oct 2016 04:17:17 +0000 (13:17 +0900)]
Update Appendix A. Release Notes.

Actual release note contents are not written yet.

9 years agoRemove Bibliography chapter.
Tatsuo Ishii [Tue, 11 Oct 2016 04:06:31 +0000 (13:06 +0900)]
Remove Bibliography chapter.

This was copied from PostgreSQL and Pgpool-II does not need it.x

9 years agoAdd "Architectural Functionality" section to Tutorial.
Tatsuo Ishii [Mon, 10 Oct 2016 22:58:55 +0000 (07:58 +0900)]
Add "Architectural Functionality" section to Tutorial.

9 years agoSGML docs for in memory query cache
Muhammad Usama [Mon, 10 Oct 2016 19:38:44 +0000 (00:38 +0500)]
SGML docs for in memory query cache

9 years agoWriting SGML docs for server administration
Muhammad Usama [Sun, 9 Oct 2016 19:14:42 +0000 (00:14 +0500)]
Writing SGML docs for server administration

9 years agoWrite tutorial.
Tatsuo Ishii [Fri, 7 Oct 2016 09:26:51 +0000 (18:26 +0900)]
Write tutorial.

9 years agoFix SGML compile errors.
Tatsuo Ishii [Thu, 6 Oct 2016 21:58:03 +0000 (06:58 +0900)]
Fix SGML compile errors.

openjade  -wall -wno-unused-param -wno-empty -wfully-tagged   -d stylesheet.dsl -t sgml -i output-html  -i include-index pgpool.sgml
openjade:loadbalance.sgml:28:31:E: character "_" is not allowed in the value of attribute "ID"
openjade:healthcheck.sgml:62:31:E: character "_" is not allowed in the value of attribute "ID"
openjade:healthcheck.sgml:89:33:E: character "_" is not allowed in the value of attribute "LINKEND"
openjade:failover.sgml:191:25:E: character "_" is not allowed in the value of attribute "ID"

9 years agoWriting SGML docs for SSL config
Muhammad Usama [Thu, 6 Oct 2016 19:39:02 +0000 (00:39 +0500)]
Writing SGML docs for SSL config

9 years agoWriting some more SGML docs
Muhammad Usama [Thu, 6 Oct 2016 16:32:50 +0000 (21:32 +0500)]
Writing some more SGML docs

9 years agoAdd SGML docs for load balancing config
Muhammad Usama [Thu, 6 Oct 2016 15:23:01 +0000 (20:23 +0500)]
Add SGML docs for load balancing config

9 years agoAdd SGML docs for watchdog architecture
Muhammad Usama [Thu, 6 Oct 2016 09:53:18 +0000 (14:53 +0500)]
Add SGML docs for watchdog architecture

9 years agoWrite some watchdog SGML documents.
Muhammad Usama [Wed, 5 Oct 2016 12:38:28 +0000 (17:38 +0500)]
Write some watchdog SGML documents.

9 years agoadd SGML docs about authentication
Muhammad Usama [Tue, 4 Oct 2016 14:53:22 +0000 (19:53 +0500)]
add SGML docs about authentication

9 years agoadd SGML docs about health check
Muhammad Usama [Tue, 4 Oct 2016 13:27:18 +0000 (18:27 +0500)]
add SGML docs about health check

9 years agoadd SGML docs of watchdog tutorial
Muhammad Usama [Tue, 4 Oct 2016 12:59:02 +0000 (17:59 +0500)]
add SGML docs of watchdog tutorial

9 years agoadd sgml docs
pengbo [Thu, 29 Sep 2016 08:28:41 +0000 (17:28 +0900)]
add sgml docs

9 years agoadd SGML docs of pg_md5
pengbo [Thu, 29 Sep 2016 07:58:49 +0000 (16:58 +0900)]
add SGML docs of pg_md5

9 years agoAdd "Setting up pgpool.conf" section to installation.sgml
Tatsuo Ishii [Wed, 28 Sep 2016 09:00:27 +0000 (18:00 +0900)]
Add  "Setting up pgpool.conf" section to installation.sgml

Still need more work in this section...

9 years agoAdd restrictions section.
Tatsuo Ishii [Wed, 28 Sep 2016 06:40:38 +0000 (15:40 +0900)]
Add restrictions section.

9 years agoFix resource leak.
Muhammad Usama [Tue, 27 Sep 2016 15:30:49 +0000 (20:30 +0500)]
Fix resource leak.

Per coverity CID 1355044.

9 years agoadd SGML doc of pcp command
pengbo [Tue, 27 Sep 2016 08:49:19 +0000 (17:49 +0900)]
add SGML doc of pcp command

9 years agoChange the default value of search_primary_node_timeout from 10 to 300.
Tatsuo Ishii [Sat, 24 Sep 2016 01:42:12 +0000 (10:42 +0900)]
Change the default value of search_primary_node_timeout from 10 to 300.

Prior default value 10 seconds is sometimes too short for a standby to
be promoted.

Per [pgpool-general: 5026].

9 years agoFix the case when all backends are down then 1 node attached.
Tatsuo Ishii [Wed, 21 Sep 2016 00:26:46 +0000 (09:26 +0900)]
Fix the case when all backends are down then 1 node attached.

When all backends are down, no connection is accepted. Then 1
PostgreSQL becomes up, and attach the node using pcp_attach_node. It
successfully finishes. However, when a new connection arrives, still
the connection is refused because pgpool child process looks into the
cached status, in which the recovered node is still in down status if
mode is streaming replication mode (native replication and other modes
are fine). Solution is, if all nodes are down, force to restart all
pgpool child.

Per bug 248.

9 years agoFix for: [pgpool-general: 4997] Avoiding downtime when pgpool changes require a restart
Muhammad Usama [Mon, 19 Sep 2016 20:22:49 +0000 (01:22 +0500)]
Fix for: [pgpool-general: 4997] Avoiding downtime when pgpool changes require a restart

To fix this, The verification mechanism of configuration parameter values is
reversed, previously the standby nodes used to verify their parameter values
against the respective values on the master pgpool-II node and when the
inconsistency was found the FATAL error was thrown, now with this commit the
verification responsibility is delegated to the master pgpool-II node.
Now the master node will verify the configuration parameter values of each
joining standby node against its local values and will produce
a WARNING message instead of an error in case of a difference.
This way the nodes having the different configurations will also be allowed to
join the watchdog cluster and the user has to manually look out for the
configuration inconsistency warnings in the master pgpool-II log to avoid the
surprises at the time of pgpool-II master switch over.

9 years agoFixing a problem with the watchdog failover_command locking mechanism
Muhammad Usama [Mon, 19 Sep 2016 19:38:49 +0000 (00:38 +0500)]
Fixing a problem with the watchdog failover_command locking mechanism

From pgpool-II 3.5 watchdog was using the separate individual locks for each
node-failover command(failover, failback and follow-master) and the lock was
acquired just before executing the respective failover script and was released
as soon as the script execution finishes. This technique although was very
efficient but also had a problem. If the failover command takes a very little
time and gets finished before the lock request from other pgpool-II node
arrives, the other node is also granted a lock, since the lock was already
released by the first node at that time. Consequently, both nodes ends up
executing the failover script.
So to fix this we are reverting back to the tested failover interlocking design
used prior to pgpool-II 3.5 where all the commands gets locked at the failover
start by the node that becomes a lock-holder and each command lock is released
after its execution finishes. And only the lock-holder node is allowed to
acquire/release the individual command lock. That way the lock-holder node
keeps the lock-holder status throughout the span of the failover execution and
the system becomes less time sensitive.

The issue was identified by Yugo<nagata@sraoss.co.jp>

9 years agoadd SGML doc [pcp_node_info]
pengbo [Fri, 16 Sep 2016 06:45:07 +0000 (15:45 +0900)]
add SGML doc [pcp_node_info]

9 years agofix some mistakes by importing 9.6 parser
pengbo [Fri, 16 Sep 2016 01:23:24 +0000 (10:23 +0900)]
fix some mistakes by importing 9.6 parser

9 years agoWrite some SGML documents.
Tatsuo Ishii [Tue, 13 Sep 2016 09:01:30 +0000 (18:01 +0900)]
Write some SGML documents.

9 years agoFix wrong printf format specification.
Tatsuo Ishii [Tue, 13 Sep 2016 05:15:57 +0000 (14:15 +0900)]
Fix wrong printf format specification.

Per coverity CID 1362585.

9 years agoDisable strict aliasing optimization.
Tatsuo Ishii [Mon, 12 Sep 2016 22:57:21 +0000 (07:57 +0900)]
Disable strict aliasing optimization.

flatten_set_variable_args() was imported from PostgreSQL in Pgpool-II
3.5. To make the code work, a compiler flag "-fno-strict-aliasing" is
necessary (PostgreSQL does so). Unfortunately when the function was
imported, the compiler flag was not added. To fix this, configure.ac
was modified.

Per [pgpool-general: 4975].

9 years agoMemory leak fix
Muhammad Usama [Fri, 9 Sep 2016 18:34:30 +0000 (23:34 +0500)]
Memory leak fix

Per Coverity CID 1355044

9 years agoFixing an obvious logic mistake.
Muhammad Usama [Fri, 9 Sep 2016 18:07:57 +0000 (23:07 +0500)]
Fixing an obvious logic mistake.

per Coverity CID 1362580 and CID 1362581

9 years agofix indent
pengbo [Fri, 9 Sep 2016 06:30:42 +0000 (15:30 +0900)]
fix indent

9 years agoMerge remote-tracking branch 'origin/master' into IMPORT_96_PARSER
pengbo [Fri, 9 Sep 2016 04:58:53 +0000 (13:58 +0900)]
Merge remote-tracking branch 'origin/master' into IMPORT_96_PARSER

9 years agofix some mistake of COPY query in case of replication mode
pengbo [Fri, 9 Sep 2016 04:29:54 +0000 (13:29 +0900)]
fix some mistake of COPY query in case of replication mode

9 years agoMerge branch 'IMPORT_96_PARSER' of ssh://git.postgresql.org/pgpool2 into IMPORT_96_PARSER
Tatsuo Ishii [Fri, 9 Sep 2016 03:00:13 +0000 (12:00 +0900)]
Merge branch 'IMPORT_96_PARSER' of ssh://git.postgresql.org/pgpool2 into IMPORT_96_PARSER

9 years agoFix mistake in previous commit.
Tatsuo Ishii [Fri, 9 Sep 2016 02:36:11 +0000 (11:36 +0900)]
Fix mistake in previous commit.

9 years agoDo not use random() while generating MD5 salt.
Tatsuo Ishii [Fri, 9 Sep 2016 02:24:54 +0000 (11:24 +0900)]
Do not use random() while generating MD5 salt.

random() should not be used in security related applications.  To
replace random(), import PostmasterRandom() from PostgreSQL.  Also
store current time at the start up of Pgpool-II main process for later
use.

Per Coverity CID 1362583.

9 years agoRemove useless assignment.
Tatsuo Ishii [Fri, 9 Sep 2016 00:57:14 +0000 (09:57 +0900)]
Remove useless assignment.

Per coverity CID 1362589.

9 years agosome changes according to coding guidelines
pengbo [Tue, 6 Sep 2016 06:19:43 +0000 (15:19 +0900)]
some changes according to coding guidelines

9 years agoIn PostgreSQL 9.6, in COPY query copy the output of an
pengbo [Mon, 5 Sep 2016 01:23:15 +0000 (10:23 +0900)]
In PostgreSQL 9.6, in COPY query copy the output of an
INSERT/UPDATE/DELETE ... RETURNING query is allowed.

In Pgpool, all of the (COPY...TO... ) was send to either primary node
or load balance node. Sometimes it causes a error.
To fix that, the query (COPY (UPDATE | INSERT | DELETE ... RETURNING ...) TO ...)
is only sent to primary node.

9 years agoadd PostgreSQL 9.6 parser
pengbo [Fri, 26 Aug 2016 07:16:11 +0000 (16:16 +0900)]
add PostgreSQL 9.6 parser

9 years agoimport PostgreSQL 9.6 parser
Peng Bo [Fri, 26 Aug 2016 07:05:14 +0000 (16:05 +0900)]
import PostgreSQL 9.6 parser

9 years agoEnhance performance of SELECT when lots of rows involved.
Tatsuo Ishii [Tue, 6 Sep 2016 08:31:45 +0000 (17:31 +0900)]
Enhance performance of SELECT when lots of rows involved.

(From pgpool-hackers: 1784)

While taking care of bug track item 238:

http://www.pgpool.net/mantisbt/view.php?id=238

I found there's a room to enhance the performance of data transfer
from Pgpool-II to frontend. In an extreme case, such as sending many
rows to frontend, the enhancement was 47% to 62% according to my
benchmark result.

Currently Pgpool-II flushes data to network (calling write(2)) every
time it sends a row data ("Data Row" message) to frontend. For
example, if 10,000 rows needed to be transfer, 10,000 times write()s
are issued. This is pretty expensive. Since after repeating to send
row data, "Command Complete" message is sent, it's enough to issue a
write() with the command complete message. Also there are unnecessary
flushing are in handling the command complete message.

Attached patch tries to fix them. I did a quick benchmark on my laptop
(Ubuntu 14.04 LTS, 2 cores, 16GB mem, SSD) using pgbench (scale factor
= 1) + custom query (SELECT * FROM pgbench_accounts), with master
branch head and with patched version. The graph attached.

Unfortunately, performance in workloads where transferring few rows,
will not be enhanced since such rows are needed to flush to network
anyway.

9 years agosome changes according to coding guidelines
pengbo [Tue, 6 Sep 2016 06:19:43 +0000 (15:19 +0900)]
some changes according to coding guidelines

9 years agoDon't ignore sync message from frontend when query cache is enabled.
Tatsuo Ishii [Mon, 5 Sep 2016 22:35:24 +0000 (07:35 +0900)]
Don't ignore sync message from frontend when query cache is enabled.

While returning cached query result, sync message sent from frontend
is discarded. This is harmless because "ready for query" messages is
sent to frontend afterward. Problem is, AccessShareLock held by
previous parse message processing is not released until sync message
is received by the backend. Fix is, forwarding the sync message to
backend and discarding "ready for query" message returned from
backend.

Per [pgpool-hackers: 1787].

9 years agoIn PostgreSQL 9.6, in COPY query copy the output of an
pengbo [Mon, 5 Sep 2016 01:23:15 +0000 (10:23 +0900)]
In PostgreSQL 9.6, in COPY query copy the output of an
INSERT/UPDATE/DELETE ... RETURNING query is allowed.

In Pgpool, all of the (COPY...TO... ) was send to either primary node
or load balance node. Sometimes it causes a error.
To fix that, the query (COPY (UPDATE | INSERT | DELETE ... RETURNING ...) TO ...)
is only sent to primary node.

9 years agoFix for 237: Pgpool-II fails to start if listen_addresses is empty string
Muhammad Usama [Mon, 29 Aug 2016 14:32:31 +0000 (19:32 +0500)]
Fix for 237: Pgpool-II fails to start if listen_addresses is empty string

The socket descriptor array (fds[]) was not getting the array end marker
when TCP listen addresses are not used.

9 years agoAdd comment to the document about connection_cache.
Tatsuo Ishii [Sat, 27 Aug 2016 08:22:26 +0000 (17:22 +0900)]
Add comment to the document about connection_cache.

Mention that connections to template0, template1, postgres and
regression databases are not cached even if connection_cache is on.

9 years agoadd PostgreSQL 9.6 parser
pengbo [Fri, 26 Aug 2016 07:16:11 +0000 (16:16 +0900)]
add PostgreSQL 9.6 parser

9 years agoimport PostgreSQL 9.6 parser
Peng Bo [Fri, 26 Aug 2016 07:05:14 +0000 (16:05 +0900)]
import PostgreSQL 9.6 parser

9 years agoMake authentication error message more user friendly.
Tatsuo Ishii [Wed, 24 Aug 2016 03:01:52 +0000 (12:01 +0900)]
Make authentication error message more user friendly.

When attempt to connect to backend (including health checking),
currently we emit something like:

"failed to authenticate"
"invalid authentication message response type, Expecting 'R' and received '%c'"

This is too cryptic for users. So emit the error message returned from
backend (for example, "sorry, too many clients already" instead of
this. By this, user could have cleaner idea on what's going on.

Since this is not a bug fix, committed in master branch only for now.

9 years agoTighten up health check timer expired condition in pool_check_fd()
Muhammad Usama [Mon, 22 Aug 2016 13:48:48 +0000 (18:48 +0500)]
Tighten up health check timer expired condition in pool_check_fd()

check if the signal was actually the health check timer expire to make sure that
we do not declare the timer expire due to some other signal arrived while
waiting for data for health check in pool_check_fd()

9 years agoAdd Japanse SGML documentation directory doc.ja.
Tatsuo Ishii [Mon, 22 Aug 2016 04:37:14 +0000 (13:37 +0900)]
Add Japanse SGML documentation directory doc.ja.

doc.ja/src/sgml/Makefile.am and doc.ja/src/sgml/stylesheet.dsl are
changed from English version to allow to handle UTF-8.